124 3. COMPUTATIONAL METHODS FOR THE RELIABILITY OF A COMPONENT
is limit state function can be a linear function and a nonlinear function of all random variables.
If all random variables X
i
.i D 1; 2; : : : ; n/ in the limit state function are statistically indepen-
dent normal distributions, we can convert them into statistically independent standard normally
distributed random variables Z
i
.i D 1; 2; : : : ; n/ by the following equation:
Z
i
D
X
i
X
i
X
i
i D 1; 2; : : : ; n; (3.22)
where
X
i
and
X
i
are the mean and standard deviation of a normally distributed random
variable X
i
. Z
i
is a corresponding standard normal distributed random variable of the normal
distributed random variable X
i
. After these conversions, the surface of the limit state func-
tion (3.21) becomes:
g
.
S; Q
/
D S Q D g
.
X
1
; X
2
; : : : ; X
n
/
D g
1
.
Z
1
; Z
2
; : : : ; Z
n
/
D 0: (3.23)
g
1
.
Z
1
; Z
2
; : : : ; Z
n
/
D 0 is a surface specified in the standard normally distributed space (a
coordinate system), which consists of standard normally distributed random variables Z
i
.i D
1; 2; : : : ; n/.
e reliability index ˇ is the shortest distance [1, 2] from the origin of the standard normally
distributed space to the surface of the limit state function: g
1
.
Z
1
; Z
2
; : : : ; Z
n
/
D 0.
Example 3.8
e limit state function of a component is g
.
S; Q
/
D S Q. Both S and Q are normal dis-
tributions. e component strength index S has a mean
S
and a standard deviation
S
. e
component stress index Q has a mean
Q
and a standard deviation
Q
. Use this limit state func-
tion to verify that the shortest distance between the origin of the standard normally distributed
space and the surface of the limit state function g
.
S; Q
/
D S Q D 0 is the reliability index
ˇ D .
S
Q
/=
q
2
S
C
2
Q
.
Solution:
e surface of the limit state function g
.
S; Q
/
D S Q is:
g
.
S; Q
/
D S Q D 0: (a)
In the S vs. Q coordinate system where the horizontal axis is Q, and the vertical axis is S , the
limit state function will be a straight line, as shown in Figure 3.4. e region on the left side of
the line of the limit state function is safe because of S > Q. e region on the right side of the
line of the limit state function is a failure region because of S < Q.
Let us convert the component strength index S and the component stress index Q into
standard normally distributed random variables Z
S
and Z
Q
:
Z
S
D
S
S
S
; Z
Q
D
Q
Q
Q
: (b)